Hi all,
Right now I only have access to one dataset here at Columbia - the Columbia University Medical Center dataset. Is there a way to send my cohort criteria to other institutions that use OHDSI and get access to a larger sample? It looked like that was a possibility or was in the pipeline.
If you have defined your cohort definition in Atlas, and the other institutions have Atlas installed, you can export your cohort definition as a text file (as a JSON document) and send it to them electronically. They can take the text content and import it into their own Atlas instance, and generate cohorts. They can also run with ‘generate features’ so you can see the characteristics of the cohort.
-Chris
Thank you so much for your help! I hope this is not too elementary but what is the difference between a cohort and concept set? For an example, with the drug phenytoin, is it better to create a cohort or a concept set?
A concept set is a list of CDM Vocabulary concepts. You might refer to them as a ‘code list’. You might have seen in literature that the list of codes used to identify patients is written as ‘ICD9 codes for Cardiac dysrhythmias: 427.x’. That’s a sort of 'expression that you interpret to mean all the ICD9 codes that fall under the 427 group, like 427.31, 427.32, etc.
A concept set expression is a CDM vocabulary-oriented construct where you list the set of concept IDs that you want to bring in to a code list, and indicate if you should include the concepts below the concept in the hierarchy. Example: Include Depression and descendants, exclude Bipolar and descendants. The expression, much like the 427.x expression above, resolves into a larger, complete set of concepts/codes that are identified with the concept set expression.This result is the ‘concept set’.
A cohort is a group of people that satisfy a cohort expression. Example expression: Newly diagnosed depression (not including bipolar)* with no prior antidepressant exposure**. Cohort exits when person receives first antidepressant** after cohort entry… New diagnosis means ‘first diagnosis in history’. the * represents a concept set identifying those codes which are used to identify people with any depression code, not including bipolar codes.the ** represents a concept set which identifies the drug concepts which are antidepressants.
Things to note about cohort expressions: it includes logic about which codes should be used and how these codes should be found relative to the diagnosis/exposure dates. A cohort is the set of people that satisfy this criteria, and each person has a start date (when they first meet the criteria) and an end date (when they meet the criteria to exit the cohort).
So, to summarize: concept sets identify codes, cohorts identify people.
-Chris
.
Depends what you want. If you want to know what drugs are found in the vocabulary under the ingredient concept ‘phenytoin’, use a concept set expression to get the concept set. if you want to know which people have received the drug in your database, use a cohort expression to get the cohort.
Hi @nbareis, I work at Columbia on the OHDSI project and would be happy to help out with any questions or assistance you may need. Do you have access to Columbia’s instance of ATLAS? If you need help using it–to establish your desired cohort or for any part of the process–please let me know. If you already have established your cohort definition in ATLAS, you can post your study question to the researchers category of this forum to engage people at other institutions who would be willing to run your study on their own databases.